Forest Ecology and Management 259 (2009) 143–150
Contents lists available at ScienceDirect
Forest Ecology and Management
journal homepage: www.elsevier.com/locate/foreco
Frequency and characteristics of sediment delivery pathways from forest harvest
units to streams
S.E. Litschert a,*, L.H. MacDonald b
a
b
Human Dimensions of Natural Resources, Colorado State University, Fort Collins, CO 80523-1480, United States
Department of Forest, Rangeland, and Watershed Stewardship, Colorado State University, CO 80523-1472, United States
A R T I C L E I N F O
A B S T R A C T
Article history:
Received 21 June 2009
Received in revised form 21 September 2009
Accepted 23 September 2009
Timber harvest is typically the largest area of anthropogenic disturbance in forested watersheds, and
harvested areas may generate from one to five times more erosion than undisturbed areas (Motha et al.,
2003). When sediment from harvested areas reaches stream channels it can degrade water quality and
aquatic habitat. Streamside management zones (SMZs) are often prescribed to minimize sediment
delivery, but there is little information about sediment delivery through these zones. Hence the
objectives of this study were to: (1) determine the frequency of sediment delivery pathways (‘‘features’’)
from timber harvest units; (2) measure the physical characteristics and connectivity of these features;
and (3) develop models to predict the length and connectivity of features from harvest units to streams.
Nearly 200 harvest units with streamside management zones were assessed on four National Forests
in the Sierra Nevada and Cascade Mountains of California. Only nineteen features were found below
harvest units ranging in age from 2 to 18 years. Feature lengths ranged from 10 m to 220 m, and the
length was significantly related to mean annual precipitation, cosine of the aspect, elevation, and
hillslope gradient (R2 = 64%, p = 0.004). Six of the nineteen features were connected to streams and five of
the six connected features originated from skid trails. The results indicate that timber harvest alone
rarely initiated large amounts of runoff and surface erosion, particularly when newer harvest practices
were utilized. Sediment delivery from timber harvest may be further reduced by locating skid trails away
from streams, maintaining high surface roughness downslope of water bars, and promptly
decommissioning skid trails following harvest.
ß 2009 Elsevier B.V. All rights reserved.
Keywords:
Sediment
Connectivity
Timber harvest
Skid trails
Streamside
Management zones
California
1. Introduction
Anthropogenic sediment sources on forested hillslopes include
roads, skid trails, and timber harvest units (e.g., Megahan, 1972;
Beschta, 1978; Croke et al., 1999; Barrett and Conroy, 2001; Motha
et al., 2003). Most recent research has focused on roads (e.g., Luce
and Black, 1999; Jones et al., 2000; Lane and Sheridan, 2002; Coe,
2006), but timber harvest units represent the largest areas of
anthropogenic disturbance and can increase erosion rates by one to
five times relative to undisturbed areas (Motha et al., 2003).
The delivery of overland flow and sediment from disturbed
hillslopes contributes to cumulative effects such as an increase in
the size of peak flows (e.g., Jones, 2000; MacDonald and Stednick,
2003), the alteration of channel morphology (Troendle and Olsen,
1994; Madej and Ozaki, 1998), degradation of aquatic habitat
(Shaw and Richardson, 2001), reductions in reservoir storage, and
increases in pollutant transport (EPA, 2003). The delivery of
* Corresponding author. Tel.: +1 970 491 4180; fax: +1 970 491 2255.
E-mail address: [email protected] (S.E. Litschert).
0378-1127/$ – see front matter ß 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.foreco.2009.09.038
sediment from hillslope sources to the stream network and the
watershed outlet has been defined as sediment connectivity
(Bracken and Croke, 2007). Connectivity of hillslope sediment
pathways can be in the form of sediment plumes when there is an
excess of sediment relative to overland flow, or in the form of rills
and gullies when the transport capacity is greater than the
sediment load.
For this paper a rill was defined by having incised banks with no
minimum depth. Gullies were defined as incised features greater
than 30 cm deep or with a cross-sectional area larger than about
1000 cm2. Rills, gullies, and sediment plumes are collectively
described as features in this paper, and we use the term rills for
both rills and gullies unless otherwise specified.
In recent years forest management techniques have been
modified to minimize surface runoff, erosion, and connectivity.
Skid trails are often designed to follow hillslope contours
(Kreutzweiser and Capell, 2001). Streamside management zones
(SMZs) are intended to provide vegetative roughness and maintain
infiltration rates that can slow or absorb overland flow and filter
sediment out of the overland flow before it reaches a stream or
water body (Kreutzweiser and Capell, 2001; Hairsine et al., 2002).
144
S.E. Litschert, L.H. MacDonald / Forest Ecology and Management 259 (2009) 143–150
Features can penetrate SMZs and thereby connect forest harvest
units to streams (Lacey, 2000; Rivenbark and Jackson, 2004). In the
Georgia Piedmont, USA, there was an average of one feature for
every 20 acres of clearcut and site prepared land, and these were
associated with convergent topography, steeper slopes, larger
contributing areas, and less ground cover (Rivenbark and Jackson,
2004). In Australia 10-m wide buffer zones reduced the delivery of
sediment from skid trails to streams by 95% (Lacey, 2000).
Measuring and modeling connectivity and sediment delivery
are critical for quantifying and predicting the cumulative effects of
timber harvest activities in forested watersheds (Bracken and
Croke, 2007). The fieldwork described in this paper evaluated
whether the areas disturbed by timber harvest and skid trails are
connected to stream channels by rills and sediment plumes. The
specific objectives were to: (1) determine the proportion of timber
harvest units that generated rills or sediment plumes that
penetrated at least 10 m into an adjacent SMZ; (2) measure the
site characteristics, size, and connectivity of rills and plumes to
stream channels; and (3) develop models to predict the length and
connectivity of rills and sediment plumes originating from timber
harvest units.
precipitation falls mostly as snow (USDA Forest Service, 1986).
Summer convective thunderstorms are more frequent on the east
side of the mountains than on the west side (USDA Forest Service,
1983).
Forests on the west side are composed primarily of ponderosa
pine (Pinus ponderosa), sugar pine (Pinus lambertiana), Douglas fir
(Pseudotsuga menziesii), red fir (Abies magnifica), white fir (Abies
concolor), and incense cedar (Libocedrus decurrens) (USDA Forest
Service, 1983, 1986, 2002). The dominant understory shrubs are
green leaf manzanita (Arctostaphylos patula), huckleberry oak
(Quercus vaccinifolia), and mountain whitethorn (Ceanothus
cordulatus). Forests on the drier east side consist primarily of
ponderosa pine, Jeffrey pine (Pinus jeffreyi), and white fir, with
some lodgepole pine (Pinus contorta), western juniper (Juniperus
occidentalis), and black oak (Quercus kelloggii). Most forest harvest
takes place on soils weathered from andesitic, granitic, and metasedimentary material (USDA Forest Service, 1983, 1986, 2002).
2. Study area
The study was conducted on the four NFs shown in Fig. 1
because they have higher levels of timber harvest than the other
eighteen NFs in Region 5 of the USDA Forest Service (USFS). Most of
the recent timber management projects in these NFs tend to be
larger-scale projects of about five hundred to several thousand
hectares. Individual harvest units within these project areas
average 15 ha with a general range of 1–80 ha (Tangenberg, USFS,
pers. comm., 2008).
Harvest projects with erosion and sedimentation problems
were identified by direct discussions with USFS personnel and
querying the USFS Best Management Practices Evaluation Program
(BMPEP) database (USDA Forest Service, 2004). The BMPEP
The study area included timber harvest units on the Eldorado,
Lassen, Plumas, and Tahoe National Forests (NF) in the Sierra
Nevada and Cascade mountains of California (Fig. 1). The area has a
Mediterranean climate with moist air flows from the Pacific Ocean.
The amount and type of precipitation is heavily influenced by the
high elevations and rain shadow effect of the mountains so that
mean annual precipitation ranges from as much as 2000 mm on
the west side to as little as 370 mm on the east side (Teale GIS
Solutions Group, 1997). Ninety-five percent of the precipitation
occurs during the winter wet season, and above 1500 m the
3. Methods
3.1. Data collection
Fig. 1. Location of the four National Forests used to assess harvest unit connectivity. The open diamonds indicate the location of the rills and sediment plumes identified in this
study. Given the coarse scale of this map, some diamonds represent multiple features that occurred in close proximity.
S.E. Litschert, L.H. MacDonald / Forest Ecology and Management 259 (2009) 143–150
provides 29 standardized forms for USFS personnel to evaluate the
effects of timber harvest, roads, grazing, mining, prescribed fires,
recreation, and engineering on soils, runoff, erosion and streams.
Data from about 3000 evaluations were available for this research.
This approach to identifying harvest projects was used to increase
the likelihood of finding rills and sediment plumes and thereby
obtains an adequate sample size for modeling purposes. The
resulting bias in data collection has minimal effect on the key
findings and conclusions presented here.
Maps of each timber harvest project were obtained from the
responsible NF and used to identify harvest units that were
immediately upslope of SMZs. The lower edges of the selected
harvest units were traversed on foot to identify all rills, gullies and
sediment plumes entering SMZs. USFS policy requires 90-m wide
SMZs along each side of perennial streams and 45-m wide SMZs
along each side of all ephemeral and intermittent streams.
Harvesting and machinery are not allowed within SMZs, although
we did observe some vehicle tracks from heavy equipment and
some skid trails crossing the SMZs.
A set of criteria was established to ensure that the erosional
features being measured were only due to forest harvest activities.
Post-fire salvage projects and harvest units that had burned were
excluded because the effects of burning could not be separated
from the effects of timber harvest. Features initiated by paved and
unpaved roads were excluded; features initiated by skid trails were
included. Features that ended at a road or that were extended by
road drainage were excluded because we could not determine the
length that a feature would have attained in the absence of the
road. A minimum feature length of 10 m was used because this is
the highest resolution of the digital elevation models (DEMs) being
used to model cumulative watershed effects.
When a feature was found, the following data were obtained:
years since harvest, mean annual precipitation (MAP), soil depth,
soil erodibility (K), straight line and feature lengths, feature
gradient, aspect, elevation, hillslope gradient, hillslope curvature,
surface roughness, and whether or not the feature was connected.
A feature was classified as connected if it extended to within 10 m
of a stream channel, indicating that at least some runoff and
sediment is likely to be delivered to the stream channel (Croke and
Mockler, 2001). The focus on quantifying the frequency of
connectivity between harvest units and streams does not
necessarily mean that large amounts of runoff and sediment are
being delivered or that water quality is likely to be impaired.
Years since harvest was determined from project documents or
estimated from vegetation regrowth. MAP at 127 mm intervals
was obtained from statewide data (Teale GIS Solutions Group,
1997) since not all NFs had isohyetal maps. Soil depth and
erodibility were collected from the soil surveys for each NF.
Harvest type was not available or necessarily obvious for every unit
although units older than 10 years generally were clearcut and
more recently harvested units were either thinned or subjected to
group selection. Age of harvest was a more reliable variable in the
statistical analysis than the estimated harvest type.
Feature length was measured with a flexible tape. Feature
gradient was measured in the field using a clinometer and aspect
was measured with a compass. The cosine of the aspect was used to
convert degrees from circular (0–3608) to continuous form for
statistical analysis. Elevation, hillslope gradient and curvature
were derived from 10-m digital elevation models (DEMs) obtained
from each NF. In the absence of a simple, quantitative measure,
current surface roughness was classified into one of five qualitative
categories as defined in Table 1. Current surface roughness was
classified since past conditions could not be reliably determined.
The drainage area that contributed to the feature was not
determined because in many cases the contributing area could
not be reliably identified due to the extensive regrowth, complex-
145
Table 1
Description of the five classes used to qualitatively characterize the surface
roughness along each of the rills and sediment plumes in this study.
Roughness class
Description
1
2
Bare mineral soil with little surface roughness.
Greater than 50% bare soil with no more than
1–2 short sections with rocks or slash.
Greater than 50% bare soil and more than
two short sections with rocks or slash.
Less than 50% bare soil, and more than
two short sections with rocks or slash.
Dense cover of vegetation or litter with
extensive slash, rocks, or woody debris.
3
4
5
ity of microtopography and skid trails, and subsequent disturbance
by management activities.
Each rill or sediment plume was divided into 10 m segments
beginning at the upslope end. For each segment the gradient was
measured with a clinometer and the surface roughness was
estimated following Table 1. For twelve of the fifteen rills the mean
width, mean depth, and maximum depth were measured for each
rill segment. These values were averaged to obtain a mean width,
mean depth, and mean maximum depth for each rill.
3.2. Data analysis
The dataset consisted of the independent variables characterizing the features and two response variables: feature
length and connectivity class. Two-tailed t-tests were conducted
to determine if the mean rill lengths were significantly different
from the mean sediment plume lengths, and if there was a
significant difference in mean gradients between rills and
sediment plumes. Since the lengths and gradients were not
significantly different by feature type, the features were grouped
for subsequent analyses. A log transformation of feature lengths
was used to obtain a more normal frequency distribution for
parametric statistical analysis (Ott and Longnecker, 2001).
Linear regression was used to assess the relationship between
each independent variable and the log-transformed length.
Associations between the independent variables were assessed
using Spearman’s rank correlation matrix as not all of the
variables were normally distributed (Ott and Longnecker, 2001).
Linear regression was used to determine whether the 10-m
gradients along each feature varied with distance from the top of
the feature.
The independent variables that were more strongly correlated
with feature length were then used to develop a multivariate linear
regression model to predict feature length. Model selection criteria
were the adjusted R2 and Mallow’s C(p) where jC(p) pj is closest
to zero, with p being the number of parameters including the
interception point (Ott and Longnecker, 2001). Partial R2 values
were calculated for each variable included in the models.
Connectivity was modeled using Fisher’s exact test.
4. Results
The lower boundaries of approximately 200 harvest units were
surveyed. Only nineteen features were found, and these included
fifteen rills or gullies and four sediment plumes (Table 2). Fourteen
or seven percent of the harvest units had at least one rill or
sediment plume entering the SMZ, and five of these 14 units had
two features each. Sixteen of the nineteen features originated from
skid trails.
The mean rill length was 43 m (S.D. = 54 m), and the range was
from 11 m to 220 m (Table 2). The median rill length was only 22 m
or 50% of the mean, indicating a highly skewed distribution. The
S.E. Litschert, L.H. MacDonald / Forest Ecology and Management 259 (2009) 143–150
146
Table 2
Mean and standard deviation (S.D.) of the characteristics of rills and sediment plumes. MAP is mean annual precipitation.
Feature ID
Rill or
plume
(R or P)
Total
length
(m)
Years
since
harvest
MAP
(mm)
Connected
(y/n)
Rill
gradient
(%)
Rill
roughness
Aspect
(8)
Soil
erodibility
(K)
Hillslope
gradient
(%)
Elevation
(m)
Soil
depth
(m)
Sinuosity
Alder_1
Alder_2
Alder_3
Ant_1
Ant_5
BigC10_1
Blak7_1
Cate14_1
LC18_1
Poison21_1
Sieg53_1
Spike2_1
Spike9_1
Verdi_1
Ward_1
R
R
R
R
R
R
R
R
R
R
R
R
R
R
R
80.0
68.0
220
22.0
25.0
11.8
33.0
16.0
11.0
58.0
11.6
11.8
29.6
18.3
10.8
18
18
18
2
2
11
8
9
5
5
4
3
11
10
13
1524
1270
1270
635
635
508
889
1016
508
635
508
762
762
762
1016
n
n
y
n
y
y
y
n
n
y
n
y
n
n
n
22
13
11
26
28
17
17
18
9
32
29
12
14
14
36
1
1
1
2
2
1
2
1
1
2
2
2
3
3
2
270
270
180
250
170
160
180
0
140
240
70
225
300
220
0
0.00
0.24
0.24
0.23
0.27
0.30
0.19
0.18
0.27
0.30
0.33
0.27
0.27
0.00
0.27
31
36
20
13
20
15
12
5
8
28
29
11
11
10
29
1785
1756
1750
1543
1590
1747
1815
1809
1763
1698
1787
1844
1852
1778
1831
0.51
1.14
1.14
0.74
0.88
0.87
1.41
0.76
1.46
0.87
0.47
1.08
1.08
1.10
1.06
1.20
1.41
1.31
1.07
1.10
1.11
1.07
1.03
1.05
1.12
1.05
1.16
1.06
1.10
1.08
9
6
847
314
20
8
1.7
0.7
178
93
0.22
0.10
19
10
1757
88
1.0
0.3
1.13
0.10
2
2
2
11
635
635
635
508
26
30
20
7
2
3
1
1
50
50
45
200
0.23
0.23
0.23
0.30
28
32
7
15
1546
1543
1536
1749
0.74
0.74
0.74
0.87
1.03
1.14
1.05
1.16
4
5
603
64
21
10
1.8
1.0
86
76
0.25
0.04
21
12
1594
104
0.8
0.1
1.10
0.06
Mean
S.D.
Ant_2
Ant_3
Ant_4
BigC10_2
42
54
P
P
P
P
21.7
19.1
10.0
14.3
Mean
S.D.
16
5
n
n
n
n
median sediment plume length was 17 m (S.D. = 5 m), and the
lengths varied only from 10 m to 22 m (Table 2). The high
variability in the rill lengths meant that there was no significant
difference between the lengths of the rills and the lengths of the
sediment plumes (p = 0.093).
Rills were generally small, as the median width was less than
50 cm and the median depth was only 5 cm (Table 3). The range of
widths was from 15 cm to 355 cm, and the presence of several
gullies resulted in an overall mean width of 90 cm (S.D. = 99 cm)
and a mean maximum depth of 19 cm (S.D. = 27 cm) (Table 3). Rill
depths were highly variable between and along rills so rill depth
was not significantly related to rill width or rill length.
Hillslope gradient had little relationship to whether a feature
was a rill or a sediment plume. The mean rill gradient was 20%, and
rills occurred on hillslopes with gradients ranging from 9% to 36%
(Table 2). The mean gradient for sediment plumes was 21% or
nearly identical to the mean value for rills, and the plumes also
occurred on hillslopes with a wide range of gradients (7–30%).
Feature gradients were generally similar to the hillslope gradients
(Table 2). The lack of any significant differences between the rills
and the sediment plumes in terms of their lengths, gradients, and
hillslope gradients meant that these features were lumped for
further analyses.
Log-transformed feature length increased significantly with
MAP (R2 = 0.45, p = 0.002; Fig. 2) and years since harvest (R2 = 0.34, Q1
p = 0.009). The two longest features drove both of these relationships, as they were in an area with high MAP and on 18-year old
clearcut units (Fig. 3). If these features are excluded, neither MAP
nor years since harvest were significantly related to feature length.
Feature lengths tended to increase on steeper hillslopes, but
this relationship was relatively weak (R2 = 0.15, p = 0.10). All of the
features were on hillslopes with low to moderate surface roughness (classes 1–3 in Table 1), as the mean surface roughness was
1.7–1.8 for both rills and sediment plumes (Table 2). Aspect, soil
Table 3
Mean width, mean maximum depth, and mean depth for the rills and gullies. S.D. is
standard deviation and N/A means the data are not available.
Feature name
Mean feature
width (cm)
Mean maximum
depth (cm)
Mean
depth (cm)
Alder_1
Alder_2
Alder_3
Antelope_1
Antelope_5
BigC10_1
Blake7_1
Cate14_1
LowerC18_1
Poison21_1
Siegfried53_1
Spike2_1
Spike9_1
Verdi_1_2
Ward_1
355
117
177
21
N/A
40
N/A
85
18
34
15
25
131
58
N/A
93
13
47
6
N/A
8
N/A
10
4
3
7
2
25
10
N/A
47
6
24
3
N/A
4
N/A
5
2
2
4
2
12
6
N/A
Mean
S.D.
90
99
19
27
10
13
Fig. 2. Length of each rill and sediment plume plotted against mean annual
precipitation.
S.E. Litschert, L.H. MacDonald / Forest Ecology and Management 259 (2009) 143–150
147
Fig. 3. Feature locations and mean annual precipitation. Given the scale of this map, one symbol may represent multiple features if they were all on the same or adjacent
harvest units.
erodibility, soil depth, and the other independent variables were
not significantly related to feature length.
The strongest correlations amongst the independent variables
were the inverse relationship between MAP and soil erodibility
(r = 0.58, p = 0.01), the inverse relationship between cosine
(aspect) and soil depth (r = 0.55, p = 0.01), and the positive
relationship between MAP and years since harvest (r = 0.54,
p = 0.02) (Table 4). The inverse relationship between MAP and
soil erodibility may be due to more dense vegetation in areas with
higher MAP, as this would increase soil organic matter and
permeability, and decrease surface runoff and erosion. The
negative relationship between aspect and soil depth indicates
that soils tended to be deeper on north-facing slopes.
The relationship between MAP and years since harvest was
driven by the two longest features as noted above. MAP was
weakly correlated with elevation for the fourteen sites
with features (r = 0.44, p = 0.06) (Table 4). This weak relationship can be attributed to the confounding effect of the rain
shadow on the eastern side of the Sierra Nevada, the relatively
small range of elevations where the features were found (1536–
1852 m), and the potential for the features to be formed by
convective storms.
Table 4
Correlation matrix for the predictor variables with r values on top and p-values below. Significant relationships (p < 0.05) are in bold.
Age of harvest
Precipitation
Elevation
Slope
Soil depth
Soil erodibility
Cosine (aspect)
Mean annual precipitation
Elevation
Cosine (aspect)
Rill roughness class
0.54
0.02
0.51
0.02
Slope
0.21
0.40
Soil depth
0.38
0.11
Soil erodibility
0.02
0.95
0.11
0.65
0.39
0.10
0.44
0.06
0.15
0.54
0.26
0.27
0.58
0.01
0.08
0.74
0.05
0.83
0.16
0.51
0.40
0.09
0.04
0.86
0.04
0.87
0.13
0.60
0.28
0.25
0.18
0.46
0.13
0.60
0.09
0.72
0.04
0.86
0.55
0.01
0.02
0.93
0.17
0.49
0.05
0.84
0.08
0.75
S.E. Litschert, L.H. MacDonald / Forest Ecology and Management 259 (2009) 143–150
148
Fig. 4. Average hillslope gradients decreased with increasing distance from the top
of the rills (R2 = 0.54, p = 0.04). The bars show one standard deviation and the values
below each bar represent the number of rills.
All of the independent variables that had at least a weak
relationship with feature length were included in the multivariate
model development. The best multivariate model for feature
length (L) included MAP, hillslope gradient (S in m m1), cosine of
the aspect (cos A), and elevation (E in m):
log L ¼ 1:852 þ 0:0224 MAP þ 0:0104 S 0:2419 cos A
0:0007 E
2
(1)
This model had a R of 0.64, a Mallow’s C(p) of 0.06, and it was
significant at p = 0.004. Partial R2 values show that MAP explained
45% of the variability in feature length, followed by 9% for hillslope
gradient, 7% for cosine (aspect), and 4% for elevation. The model
tended to over-predict the length of shorter features and underpredict the length of longer features.
Hillslope gradients tended to decrease with distance from the
top of the rills (R2 = 54%; p = 0.03; Fig. 4). Rill width and rill depth
did not show a strong, consistent relationship with the absolute or
relative distance from the top of each rill. In some rills rocks, roots,
or trees affected the size and shape of the rill, and the resulting
abrupt changes in gradient and channel dimensions helped explain
the lack of any consistent relationships between rill dimensions
and the distance from the top of each rill.
Six of the fifteen rills and none of the sediment plumes were
directly connected to the stream (Table 3). Five of the six connected
rills originated from skid trails and the other connected rill
originated from a clearcut. The connected rills ranged in length
from 12 m to 220 m (Fig. 5), and the short lengths of four of the
connected features indicate that these originated from skid trails in
the SMZ.
Connectivity was not related to rill or hillslope gradient, as the
connected rills were found on slopes ranging from 11% to 32%. The
Fig. 5. Frequency and connectivity of features by length class.
average length of connected features was 60 m as compared to
26 m for the unconnected features, but if the longest connected rill
(220 m) is excluded, the mean length of the connected rills drops to
28 m, or nearly the same mean length as the unconnected features.
Surprisingly, the connected features tended to occur in areas with
lower MAP, as the MAP for the connected features was only
660 mm as compared to the MAP of 910 mm for the unconnected
features, although this difference was not significant. Univariate
analyses showed only weak relationships between connectivity
and the independent variables of time since disturbance, soil
erodibility, soil depth, elevation, and aspect. Efforts to predict
connectivity by feature length, slope or other variables using
Fisher’s exact test were unsuccessful.
5. Discussion
The source of each rill or sediment plume was either a water bar
on a skid trail (n = 16) or a clearcut (n = 3), and this provides both
insights into the causative processes and management implications. Like roads, skid trails are compacted relative to the adjacent
areas, and the lower infiltration rate means that overland flow can
be readily generated (Croke et al., 1999; Croke and Mockler, 2001;
Ziegler et al., 2000; Lacey, 2000). This overland flow will increase
erosion and sediment transport rates (Luce and Black, 1999; Coe,
2006), which is why water bars are required on steeper skid trails.
More frequent water bars may be needed to reduce the area
draining to each water bar. The resulting reduction in overland
flow should reduce sediment production, and the likelihood of
sediment being delivered to streams (Croke and Mockler, 2001;
Coe, 2006).
Our field observations and the data both indicate that the
concentrated flow from a skid trail and water bar was more likely
to form a rill or sediment plume when the downslope area had low
surface roughness. In some cases, tractor tracks or logs provided
enough surface roughness to stop a rill from becoming deeper or
longer. Sediment travel distance below water bars also can be
reduced by the deposition and compaction of litter or slash to
increase surface roughness (Ketcheson and Megahan, 1996). Key
management implications are that the delivery of runoff and
sediment to streams can be minimized by keeping skid trails out of
the SMZ, increasing the frequency of water bars, and maximizing
surface roughness.
Areas that have been clearcut with ground-based logging
equipment also can generate overland flow and surface erosion.
In recent years, the amount of clearcutting on federal lands has
been greatly reduced due to the potential for adverse
consequences on soils, water, plant and animal diversity,
aesthetics, and recreation (Backiel and Gorte, 1992). A selection
cut or thinning typically disturbs a smaller proportion of the
harvest unit, and this will reduce the likelihood of overland flow
and rill initiation. It is noteworthy that two of the three longest
gullies came from an 18-year old clearcut with highly erodible,
coarse-textured, granitic soils, a sparse grass cover with low
surface roughness, and rocky outcrops that would readily
generate infiltration-excess overland flow. The Forest Service
has since mitigated the erosion at this site, and the implication
is that older clearcuts should be checked for legacy features that
may be generating and delivering sediment to streams. Such
sites may need additional treatments to minimize surface
runoff, surface erosion, and sediment delivery.
One of the more remarkable findings from this study is the
small number of features identified by the field inspection of 200
harvest units. The increasing use and refinement of Best Management Practices (BMPs), plus the shift from clearcuts to thinning
and group selection, can help explain the relative lack of rills and
sediment plumes from harvest units on these four NFs.
S.E. Litschert, L.H. MacDonald / Forest Ecology and Management 259 (2009) 143–150
Three of the four NFs in this study also fall within the area
regulated by the 1998 Herger-Feinstein Quincy Library Group
(HFQLG) Forest Recovery Act. This Act requires the retention of at
least 50% of the forest canopy in the units being harvested, so the
specified annual harvest of 13,000–29,000 ha (2–4% of the total
Pilot Project area) is occurring through group selection,
individual tree selection, or thinning (HFQLG Pilot Project
Implementation Team, 2007). Both hydrologic modeling and
paired watershed studies indicate that this type and intensity of
timber harvest will have little or no effect on runoff at the
watershed scale (Stednick, 1996; HFQLG Pilot Project Implementation Team, 2007). Since the remaining trees provide
canopy cover, ground cover, and surface roughness, it should
not be surprising that 84% of the rills and sediment plumes
observed in this study originated from skid trails.
Most of the rills and sediment plumes were found on harvest
units on the east side of the four NFs examined in this study (Fig. 3).
These areas generally have lower MAP, but they are subject to
higher-intensity summer convective storms (USDA Forest Service,
1983). Qualitative field observations indicated that the east-side
harvest units had much less litter and ground cover than the
wetter, west-side units. The multivariate model for feature length
provides further evidence for the importance of surface cover, as
the negative coefficient for cosine (aspect) indicates that the longer
rills are associated with south-facing slopes. In this region the
south-facing slopes are hotter and drier during the summer, and
typically have less vegetation and litter than the sites with a north,
east, or west aspect (P. Stancheff, Plumas National Forest, pers.
comm., 2005). Less surface cover means that south-facing slopes
are more susceptible to rainsplash and surface sealing (McIntyre,
1958; Fox and Le Bissonnais, 1998; Assouline, 2004; Larsen et al.,
2009) and hence are more likely to generate infiltration-excess
overland flow. A lower amount of surface cover also will reduce
surface roughness, so runoff velocities will be higher and they will
be more prone to surface erosion.
One caveat to the results of this study is the potential for
underestimating the frequency of rills and sediment plumes as
sites recover. In the Georgia Piedmont, USA, gullies associated with
timber harvest filled in over one to two seasons (Rivenbark and
Jackson, 2004). In Australia, skid trail erosion rates decreased by an
order of magnitude by the fifth year after harvesting (Croke et al.,
2001), and this would reduce the formation and persistence of rills
and sediment plumes. To more accurately characterize and
understand the connectivity of harvest units to streams, studies
should begin immediately after harvest. Initiating a study
immediately after harvest would make it easier to define the
contributing area using either high resolution DEMs or field
surveys. At least in our area, field studies also are necessary to
determine the relative importance of rock outcrops, as these may
be even more important for rill formation than contributing area.
These types of field studies should continue for several years in
order to account for the effect of more extreme storm events.
Studies over multiple years also will allow an evaluation of the rate
at which rills and plumes recover over time and the effect of
dynamic factors, such as vegetation regrowth, on feature development, length, and connectivity (Dudziak, 1974). This longerterm monitoring is essential for understanding the extent to which
rills and sediment plumes can deliver runoff and sediment from
forest harvest units to streams.
6. Conclusions
This study investigated the frequency of rills and sediment
plumes emanating from timber harvest units and their connectivity to streams. The downslope edges of approximately two
hundred harvest units on four National Forests were traversed
149
in the Sierra Nevada and Cascade mountains of California. A total of
fifteen rills and four sediment plumes were found downslope of
fourteen timber harvest units. The mean length was 36 m, and the
maximum feature length of 220 m occurred in an 18-year old
clearcut with low surface roughness. Only six of the rills and none
of the sediment plumes were connected to streams.
Sixteen of the nineteen features originated from skid trails, and
the other three features originated in older clearcuts with very
coarse soils and sparse vegetative cover. The majority of the
features were found in drier areas, and this suggests that surface
cover and surface roughness are important controls on the
development of these features. Multivariate analysis showed that
feature length increased with mean annual precipitation, years
since harvest, cosine (aspect), and hillslope gradient (R2 = 0.64).
The downslope progression of features was often stopped by the
presence of organic matter and surface roughness such as litter,
logging slash, woody debris and bumpy microtopography.
The results indicate that the proper construction and postharvest treatment of skid trails is critical for reducing the delivery
of concentrated flow and sediment from timber harvest units to
streams. The likelihood of sediment delivery from harvested areas
can be greatly reduced by constructing more water bars along the
skid trails to reduce the amount of concentrated flow at any one
point; ripping the skid trail after harvesting to maximize
infiltration; and ensuring that the hillslope below the water bar
has as much surface roughness as possible. The limited number of
features found in this study suggests that current forest harvest
procedures and Best Management Practices are largely effective in
reducing rilling and sediment delivery.
Acknowledgments
The authors are grateful to Brian Staab and Region 5 of the US
Forest Service for providing the funding for this research. We also
want to thank Bob Schultz, Christine Mai, Chuck Mitchell, and all of
the other Forest Service personnel who provided information,
maps, and bear stories. We are also grateful to Drew Coe and the
two anonymous reviewers of this paper for their helpful
comments.
References
Assouline, S., 2004. Rainfall-induced soil surface sealing: a critical review of
observations, conceptual models, and solutions. Vadose Zone Journal 3, 570–
591.
Backiel, A., Gorte, R.W., 1992. Clearcutting in the National Forests. Congressional
Research Service (CRS) Report 92-607. http://www.ncseonline.org/NLECRSreports/Forests (accessed February 2008).
Barrett, J.C., Conroy, W.J., 2001. In: Warwick, J.J. (Ed.), Evaluation of Timber Harvest
Impacts on Water Quality, 1(1). American Water Resources Association Technical Publication Series, pp. 205–210.
Beschta, R.L., 1978. Long-term patterns of sediment production following road
construction and logging in the Oregon coastal range. Water Resources
Research 14 (6), 1011–1016.
Bracken, L., Croke, J., 2007. The concept of hydrological connectivity and its contribution to understanding runoff-dominated geomorphic systems. Hydrologic
Processes 21, 1749–1763.
Coe, D., 2006. Sediment production and delivery from forest roads in the Sierra
Nevada, CA. M.S. Thesis, Colorado State University, Fort Collins, CO, 117 pp.
Croke, J., Hairsine, P., Fogarty, P., 1999. Sediment transport, redistribution and
storage on logged forest hillslopes in south-eastern Australia. Hydrological
Processes 13 (17), 2705–2720.
Croke, J., Mockler, S., 2001. Gully initiation and road-to-stream linkage in a forested
catchment, southeastern Australia. Earth Surface Processes and Landforms 26,
205–217.
Croke, J., Hairsine, P., Fogarty, P., 2001. Soil recovery from track construction and
harvesting changes in surface infiltration, erosion and delivery rates with time.
Forest Ecology and Management 143, 3–12.
Dudziak, J., 1974. Observations of development of slope rills in forest clearings of
the Tatra Mountains (In Polish with English summary). Czasopismo Geograficzne 45 (1), 31–45.
EPA, 2003. National Water Quality Inventory, 1998 report to Congress. http://
www.epa.gov/305b/98report/ (accessed May 2006).
150
S.E. Litschert, L.H. MacDonald / Forest Ecology and Management 259 (2009) 143–150
Fox, D.M., Le Bissonnais, Y., 1998. A process-based analysis of the influence of
aggregate stability on surface crusting, infiltration, and interrill erosion. Soil
Science Society of America Journal 62, 717–724.
Hairsine, P.B., Croke, J.C., Mathews, H., Fogarty, P., Mockler, S.P., 2002. Modelling
plumes of overland flow from logging tracks. Hydrological Processes 16 (12),
2311–2327.
HFQLG Pilot Project Implementation Team, 2007. Status Report to Congress Fiscal
Year 2006. USDA Forest Service, Pacific Southwest Region, Quincy, CA, 18 pp.
Jones, J.A., 2000. Hydrologic processes and peak discharge response to forest
removal, regrowth, and roads in 10 small experimental basins, western Cascades, Oregon. Water Resources Research 36 (9), 2621–2642.
Jones, J.A., Swanson, F.J., Wemple, B.C., Snyder, K.U., 2000. Effects of roads on
hydrology, geomorphology, and disturbance patches in stream networks. Conservation Biology 14 (1), 76–85.
Ketcheson, G.L., Megahan, W.F., 1996. Sediment Production and Downslope Sediment Transport from Forest Roads in Granitic Watersheds. INT-RP-486. USDA
Forest Service, Ogden, UT, p. 11.
Kreutzweiser, D.P., Capell, S.S., 2001. Fine sediment deposition in streams after
selective forest harvesting without riparian buffers. Canadian Journal of Forest
Research 31 (12), 2134–2142.
Lacey, S.T., 2000. Runoff and sediment attenuation by undisturbed and lightly
disturbed forest buffers. Water, Air, and Soil Pollution 122 (1–2), 121–138.
Lane, P.N.J., Sheridan, G.J., 2002. Impact of an unsealed forest road stream crossing:
water quality and sediment sources. Hydrological Processes 16 (13), 2599–2612.
Larsen, I.J., MacDonald, L.H., Brown, E., Rough, D., Welsh, M.J., Pietraszek, J.,
Libohova, Z., de Dios Benavides-Solorio, J., 2009. Causes of post-fire runoff
and erosion: soil water repellency, cover or soil sealing? Soil Science Society of
America Journal 73 (4), 1303–1407.
Luce, C.H., Black, T.A., 1999. Sediment production from forest roads in western
Oregon. Water Resources Research 35 (8), 2561–2570.
MacDonald, L.H., Stednick, J.D., 2003. Forests and Water: a State-of-the-Art Review
for Colorado, CWRRI Completion Report No. 196, Colorado State University, Fort
Collins, CO.
Madej, M.A., Ozaki, V., 1998. Channel Response to Sediment Wave Propagation and
Movement, vol. 21 (10). Earth Surface Processes and Landforms, Redwood
Creek, CA, pp. 911–927.
McIntyre, D.S., 1958. Permeability measurements of soil crusts formed by raindrop
impact. Soil Science 85, 185–189.
Megahan, W., 1972. An overview of erosion and sedimentation processes on
granitic soils. In: Proceedings of the Conference on Decomposed Granitic Soils:
Problems and Solutions. University of California, Davis, Redding, CA.
Motha, J.A., Wallbrink, P.J., Hairsine, P.B., Grayson, R.B., 2003. Determining the
sources of suspended sediment in a forested catchment in southeastern Australia. Water Resources Research 39 (3) 1056, doi:10.1029/2001WR000794.
Ott, R.L., Longnecker, M., 2001. An Introduction to Statistical Methods and Data
Analysis. Duxbury Press, Pacific Grove, CA.
Rivenbark, B., Jackson, C., 2004. Concentrated flow breakthroughs moving through
silvicultural streamside management zones: Southeastern piedmont, USA.
Journal of the American Water Resources Association 40 (4), 1043–1052.
Shaw, E.A., Richardson, J.S., 2001. Direct and indirect effects of sediment pulse
duration on stream invertebrate assemblages and rainbow trout (Oncorhynchus
mykiss) growth and survival. Canadian Journal of Fisheries and Aquatic Sciences
58, 2213–2221.
Stednick, J.D., 1996. Monitoring the effects of timber harvest on annual water yield.
Journal of Hydrology 176, 79–95.
Teale GIS Solutions Group, 1997. California Annual Precipitation. URL: http://
gis.ca.gov/casil/gis.ca.gov/teale/precipa/.
Troendle, C.A., Olsen, W.K., 1994. Potential Effects of Timber Harvest and Water
Management on Streamflow Dynamics and Sediment Transport. GTR-RM-247.
USDA Forest Service, Rocky Mountain Forest and Range Experiment Station,
Fort Collins, CO, pp. 34–41.
USDA Forest Service, 2004. Best Management Practices Evaluation Program
(BMPEP) 1992–2002 Monitoring Results. USDA Forest Service, Pacific Southwest Region, Vallejo, CA.
USDA Forest Service, 1983. Plumas National Forest Soil Survey. Beckwourth, CA.
USDA Forest Service, 1986. Eldorado National Forest Soil Survey. Placerville, CA.
USDA Forest Service, 2002. Tahoe National Forest Soil Survey. Adaptive Management Services Enterprise Team, Nevada City, CA.
Ziegler, A.D., Sutherland, R.A., Giambelluca, T.W., 2000. Runoff generation and
sediment production on unpaved roads, footpaths and agricultural land surfaces in northern Thailand. Earth Surface Processes and Landforms 25 (5), 519–
534.